Low% "secret world assisted" speed run of Metroid Prime 2: Echoes in 20 segments, completed on September 22 2006.


Author's comments:

The Videos

This speed run is available in three flavours:

• "normal quality", single audio only, DivX, AVI container (the .avi files)
  
• "normal quality", dual audio with commentary track, DivX, matroska container (.mkv files)
  
• "high quality", dual audio with commentary track, DivX, matroska container (.mkv files labelled HQ)

If you wish to listen to the audio commentary track, you will need to download one of the matroska (.mkv) sets of files. Matroska was chosen for the dual audio versions of the run as multiple audio tracks in other containers is poorly supported by many video players at the present time. You are strongly advised to view the matroska files using the free, open-source, cross-platform VLC player, which can be downloaded for many operating systems at http://www.videolan.org/vlc/. To listen to the commentary in VLC, choose audio -> audio track -> 2. Additionally, if you can tolerate the substantially higher filesizes, consider downloading the high quality ("_HQ") matroska encodes, as at twice the resolution and twice the framerate they will make watching the run considerably more pleasant.

Please be aware that at the current time, archive.org's webservers are not configured to recognise mkv videos, assigning them a MIME type of text/plain. In practical terms this means that you will need to right click and choose "Download ..." or "Save as ..." etc. to download the files, rather than left-clicking on them, which will reward you with a screenful of gibberish. Standalone downloaders such as flashget, wget, downthemall and so on ought to work too.

The "PROMO_" prefixed .avi files (in normal and high qualities) are promotional music videos constructed using clips from the speedrun, and are not part of the run itself.

The run was performed on the PAL (European) version of the game, but played on an NTSC (US) gamecube using a Freeloader import enabling disc. This was done to make it possible to capture the attempt using a DVD recorder, as the PAL version only produces an unsupported PAL-60 signal. The use of the Freeloader has no effect on gameplay.

In keeping with standard procedure for recorded Echoes runs, the final segment (20) is actually two videos edited together at the Emperor Ing 3 death cutscene. For those unaware, if the player dies on the Dark Samus 3/4 fight, the game permits continuation from immediately after the death of the Emperor provided the Gamecube has not been reset (although a formal save to a memory card at this point is not allowed). Continuing from this point does not incur a time penalty, as the game timer is reset to whatever it was before the boss death cutscene. Since I needed a lucky hit on Dark Samus 3 with the Screw Attack, I committed suicide and restarted from before the fight in this way several dozen times before I was finally able to proceed to Dark Samus 4. As I explained, this has always been standard procedure on Echoes runs, but I wished to explain it explicitly for the purposes of transparency. This is the one-and-only point where any cutting or pasting of footage took place.

Skips

The "default" percentage for this game (collecting all "essential" items and none of the optional items) is 22. I skip the following "essential" items:

• emerald translator
  
• gravity boost
  
• grapple beam
  
• cobalt translator
  
• echo visor
  
• annihilator beam

Those good at arithmetic will notice that this ought to give 16% rather than 17%, but in order to make some of the above skips possible, faster or easier, I collected an early power bomb expansion in segment 7, which raises my final percentage by one.

Conception

This was not the plan.

I completed a low percent run of this game a few months after its release, achieving a time of 2:44 and a completion percentage of 20 (this abysmal run may still be found at http://www.archive.org/details/MetroidEchoes_20p_244). This old run was intended simply to put low percent "secret world-assisted" speed runs on the map as a viable category and demonstrate that it was, despite a commonly held belief to the contrary, possible to skip items outright in Echoes. I was hoping and expecting that this attempt would encourage superior players into recording further games with exponentially decreasing times and percentages, and saw my effort as simply a way to "push the boat out". I did not want to do another run on this game after my initial contribution.

While work on further item skips certainly continued apace, there seemed a disappointing lack of interest in presenting this ongoing effort in speed runs. Roughly one year after the release of Echoes, I became fed up with both the low percent video void and the dearth of concrete, formal documentation of the work that players had put in to break the game. Against my better judgement, I gradually found myself being sucked into preparing a sequel.

At this stage I was pretty clueless about what was and was not possible in terms of item skips (see the above comment about lack of documentation), and about which skips could be performed without excessive use of the notorious secret world "floating" technique, a slow-as-molasses abomination that I flatly refuse to use in a run. Silent echo soon piped up with a floatless route that would produce either a 17% or 18% game depending on whether I felt up to the recently-proven challenge of skipping the annihilator beam, something which had been proposed from the earliest days of Echoes sequence breaking but only realised roughly one year later when Master ZED unveiled a video of the skip. I began walking through silent echo's suggestion and looking for pitfalls, knowing that rushing into the old 20% without planning it out in advance had contributed significantly to its appalling quality.

This approach worked fine until I reached Grand Abyss, home of the infamous "robot hop" roll jump required to skip the Grapple Beam - hands down the hardest trick in the game (some would say the hardest trick in either of the Prime games). After many, many failures in practice, I lost interest and the speed run was relegated to the status of "possible things to do in the future if life gets really boring".

Summer arrived, and with it a selection of the metroid2002.com elite found themselves with little to do. I mentioned idly to Red Scarlet on IRC that I was considering a secret world assisted 17% attempt on Echoes, stupidly forgetting Scarlet's voracious appetite for new Metroid speed run product. From this point on any complaint about boredom was met with a gentle reminder that I had a run to do. Knowing that I had an audience meant that it didn't take much cajoling before the little black-and-blue disc was back in my Gamecube, and despite a string of technical problems that needed to be overcome to get a DVD recorder to capture a PAL-60 only game, I started the run on June 25th, 2006.

One issue that came up during the run was whether I should be doing 16% or 17%, as a floatless 16% is now believed to be possible. After agonising for a while over this choice, I eventually plumped for my original plan of 17%. There were two main reasons for this decision. Firstly the 16% game would require much more secret world use, and owing to the controversial nature of going out-of-bounds and the associated time penalty, I felt that the 17% game would be significantly more watchable than the 16%. Secondly, doing 17% made it more likely that someone else would attempt 16% later, which would mean that two runs would be produced instead of just the one, which can only be a good thing.

The big worry throughout the first three-quarters of the run was whether I would be able to make the hop to the robot in Grand Abyss. I failed well over 200 times while recording before desperation took hold and I found myself researching other possible methods. Ultimately I was able to make it onto the robot with a dash jump instead, a technique which seems much more reliable than the roll jump.

Suggestions For Improvement

This attempt contains the usual sprinkling of mistakes and sections that did not go perfectly, but there are a few things to which I would like to call particular attention. Most of them are related to lack of proper forward planning, so let these be a lesson to anyone else considering a speed run on any game.

• There are a couple of big changes that could have been made to the route. In the audio commentary I speak at length in segment 12 about the issue of "the best way to reach Sanctuary". After leaving Torvus, I could have opened up Windchamber and gone to Sanctuary Fortress via the front entrance rather than going through Agon and powerbombing a side entrance. Opening up the front entrance (which can only be done from the outside in, not from the inside out) would have given me another option to reach the Light Suit after skipping Quadraxis and would also have enabled alternative routes when the time came to collect the Sky Temple keys. My gut feeling is that I made the wrong choice and that opening up the front entrance would ultimately have been faster, but I do not think it would be possible to know for sure without running the whole second half of the game twice, which I was not prepared to do. The second change that could have been made to the route would have been to make use of a "wallcrawl" using secret world techniques from upper dark Torvus to lower dark Torvus when collecting the Sky Temple keys. I rejected this because of difficulty, apathy, and the fact that it did not seem to make sense routewise in this game, but it might have saved further time if combined with the alternative Sanctuary entrance just mentioned.
  
• Master ZED pointed out once I had finished the run that while collecting the Sky Temple keys it would have been much faster to cross Grapple Guardian's room and powerbomb a grate to reach the key in Dungeon, rather than going all the way back to Hydrodynamo Station as I did. I wasn't actually aware while doing the run that his suggested path could be taken, so I lost roughly 30 seconds here, an unpleasant omission.
  
• After skipping Quadraxis in segment 15, I use the portal in Judgement Drop to get back to the light world. It would have been probably 40 seconds quicker to use the portal in Aerial Training Site instead, but in order to do that I would have needed to open the shield covering this portal in Watch Station back in segment 13. Segment 13 was already so difficult, however, that I elected not to complicate it further, and so made the decision to trade off speed for simplicity. I do not regret this decision, and do not see it as a mistake per se, but mention it here as a means by which a bolder individual than I might cut a chunk out of the final time. Although the superficial cost is 40 seconds, it must be remembered that opening the portal would have taken some time, so the overall cost might be more like 30 seconds, although this would depend on the precise method that would be used to reach the bomb slot in Watch Station.
  
• When leaving lower Torvus for the first time, I neglected to scan the scan point in front of the light beam door, an omission that would come back to bite me when I returned to lower Torvus later to collect the main Power Bombs and pick up the Dungeon Sky Temple key. This was a real facepalm moment, but I was able to minimise the cost by making use of the "air underwater" or "screw attack underwater" glitch to reach the scan point later. This failure in forward planning probably cost between five and ten seconds. Also, when passing through Path of Eyes on the way to Torvus, I ought to have moved the light beam block protecting the Super Missile door - when I come to use this door later on to cross Windchamber during the Sky Temple key route, I am forced to wait several seconds for the block to move.
  
• While skipping Quadraxis, I fail to go out-of-bounds cleanly twice, at a cost of eight seconds. I must then recharge the energy I lost from dark world exposure during those eight seconds in the safe zone under the energy controller access room, at a further cost of ten seconds, for a total of eighteen seconds lost. This was the worst single execution mistake in the run.
  
• Most of the boss fights were pretty good, but both Emperor Ing and Chykka could have been beaten more quickly.
  
  and, of course ...
  
  
• It's not 16%! I'm sure someone out there must want to fix that situation ...

Given all of the above, plus the other mistakes and pieces of bad luck, I think it would have been theoretically possible to complete this game in 1:53; perhaps 1:52.

Thanks

Thanks to:

• nate, who was supportive and helpful in all regards, but particularly in the sphere of digital video;
  
• Radix, for allowing my evil out-of-bounds run in the speed runs section of archive.org;
  
• doninss, silent ((((echo)))), SkippyJr, and especially XtraX, Master ZED and bartendorsparky for their endlessly helpful flow of suggestions and encouragement - hope you enjoy the result, gentlemen;
  
• Red Scarlet, for bullying me into actually starting;
  
• everyone else at metroid2002.com who encouraged me and tolerated my incessant complaints about how unfair everything was;
  
• Retro Studios and Nintendo, for making the game in the first place; and
  
• the developers and maintainers of GNU/Linux, ffmpeg, mplayer/mencoder, VideoLAN, bash, gcc, mkvtoolnix, LAME, avidemux2 and audacity, without whose great free software this run would simply be a pile of DVD-RWs on a shelf.

plus-circle Add Review

comment

Reviews